As described here, many US airlines have been increasing the scheduled durations of their flights. The point is that a higher fraction of actual flights will arrive within the scheduled time, increasing the airline's "on-time" ratio and thus improving its standing with consumers. This is deceptive since, contrary to customers' expectations, on-time ratio changes while the flights themselves all take exactly the same time.
I think the best way to avoid this problem is to calculate on-time ratio in a different way. Flight time should not be compared to an arbitrary scheduled time. Rather, the FAA should take all flights between a given pair of airports in a year, and calculate the average physical time it took to fly between them (or, perhaps more usefully, the 75th percentile of flight times – the time that 75% of flights between these airports were faster than). If a given flight beats this average, it is considered "on-time". Thus, a totally objective on-time ratio can be developed.
Some tweaks may be needed to make such a formula fair, for example, comparing flights only to other flights with the same plane model or at the same time of day. But in an age when everyone and their sister are coming up with new sabermetric measurements, the difficulty of number crunching should not be an issue.